Blog Posts

10 Process Mining Talks @ BPM’09

We just returned from this year’s Business Process Management (BPM) conference in Ulm, Germany. It’s a scientific conference, but with a strong focus on practical relevance. As last year, Sandy Kemsley was live-blogging about the talks she attended. Be sure not to miss her excellent BPM2009 coverage in the Column 2 blog.

Business Process Management 2009

For me, it has been inspiring and fun to meet old and new colleagues. But most of all, the BPM (and the BPI workshop) is the platform for state-of-the-art process mining research.
4 out of 19 full papers at the BPM conference were in dedicated process mining sessions, and 6 out of the 10 BPI papers were directly related to process mining topics.
So, this post is a brief review of the 10 process mining talks this year.

07.09. 2009 – BPI workshop (see complete program)

  1. Analyzing Resource Behavior Using Process Mining presented by Joyce Nakatumba

_Process mining can help to make business process simulation more realistic by automatically extracting process characteristics from historical data. One of the remaining big challenges is the simulation of human behavior. This paper explores the effect of work load on the processing speeds of people. It is a first step to leverage event logs for the characterization of resource behavior. _

  1. Flexible Multi-Dimensional Visualization of Process Enactment Data presented by Thomas Heer
    _
    This paper uses execution traces to monitor and visualize the progress (i.e., status) of a development process in the plant engineering domain.
    While the set of measures in the data warehouse is fixed, there is a close integration of the process management system and the project status view. For example, in the presented prototype the progress of running instances can be visualized directly in the workflow designer.
    _

  2. Process Mining: Fuzzy Clustering and Performance Visualization presented by Boudewijn van Dongen
    _
    Currently, few techniques are available to project performance-related information onto discovered process models. Furthermore, events in the log may occur on different levels of abstraction.
    In the proposed approach, performance measurements are collected (and projected) depending on the chosen—possibly abstracted—process model, whereas activities can be represented by multiple clusters.
    _

  3. Understanding Spaghetti Models with Sequence Clustering for ProM presented by Diogo Ferreira
    _
    A problem often encountered in practice is that for processes with a high diversity of behavior only very complex models can be discovered. Grouping the traces into more homogeneous clusters (and discovering separate models for each of them) is one strategy to obtain better models. Here, an iterative approach based on first-order Markov Chains is used to gradually assign traces to the “best” cluster.
    _

  4. Activity Mining by Global Trace Segmentation presented by Christian W. Gnther
    _
    Trace segmentation techniques address the problem that events are often logged on a much more fine-grained level of abstraction than the tasks or activities a business analyst has in mind. To bridge this gap, activity mining groups a number of low-level events within a trace to a higher-level activity. Based on these activities, simpler process models can be discovered. Here, we present a new—global and hierarchical—activity mining approach.
    _

  5. Trace Clustering Based on Conserved Patterns: Towards Achieving Better Process Models presented by R.P. Jagadeesh Chandra Bose (JC)
    _
    Trace clustering techniques aim at the separation of groups of traces, from which more understandable process models can be discovered (compared to the mining of the whole event log). The clustering is often based on so-called features that are derived from the traces. This paper presents a new, efficient approach using pattern-based feature sets and compares the quality of the results to existing clustering approaches.
    _

09.09. 2009 – First Process Mining Session (Main Conference)

  1. Discovering Process Models from Unlabelled Event Logs presented by Diogo Ferreira (see also Sandy’s summary)
    _
    One of the fundamental assumptions in process mining is that events in the log can be associated to a particular process instance (also called case ID). This is needed to distinguish multiple, concurrent executions of the same process. Interestingly, this work focuses on finding the case ID for each event in an unlabeled event stream based on a probabilistic approach, with the goal to pre-process logs that do not fulfill this assumption.
    _

  2. Abstractions in Process Mining: A Taxonomy of Patterns presented by R.P. Jagadeesh Chandra Bose (JC)
    _
    To deal with less structured processes in real-life situations, abstractions are often needed to discover models that can be understood. Here, the manifestations of commonly used process model constructs (e.g., loops) are investigated, and the derived pattern definitions are then used as abstractions for the mining of higher-level activities in the event log—for example, as a pre-processing step for other process mining methods.
    _

10.09. 2009 – Second Process Mining Session (Main Conference)

  1. Divide-and-Conquer Strategies for Process Mining presented by Josep Carmona (see also Sandy’s summary)
    _
    The theory of regions can be used to synthesize a Petri net process model from a state space that has been derived from the event log. These techniques are attractive because it is possible to, for example, steer the degree of generalization or amount of duplicate tasks in the resulting process model. A problem is that they have a high complexity, which is addressed in this paper by decomposition and clustering techniques.
    _

  2. Discovering Reference Models by Mining Process Variants Using a Heuristic Approach presented by Chen Li (see also Sandy’s summary)
    _
    Adaptive process management systems allow for structural process changes both during design and runtime. One problem is that the provided flexibility leads to a number of process variants derived from the same model in a running system. Here, an algorithm to learn from past process changes by mining process variants is proposed to support the maintanance and configuration of such process variants.
    _

Leave a Comment

Get the BPI Web Feed

Using the HTML code below, you can display this Business Process Incubator page content with the current filter and sorting inside your web site for FREE.

Copy/Paste this code in your website html code:

<iframe src="https://www.businessprocessincubator.com/content/10-process-mining-talks-bpm09/?feed=html" frameborder="0" scrolling="auto" width="100%" height="700">

Customizing your BPI Web Feed

You can click on the Get the BPI Web Feed link on any of our page to create the best possible feed for your site. Here are a few tips to customize your BPI Web Feed.

Customizing the Content Filter
On any page, you can add filter criteria using the MORE FILTERS interface:

Customizing the Content Filter

Customizing the Content Sorting
Clicking on the sorting options will also change the way your BPI Web Feed will be ordered on your site:

Get the BPI Web Feed

Some integration examples

BPMN.org

XPDL.org

×